SkillAgentSearch skills...

setup

Configure claude-starter's pieces that a plugin can't ship itself - the optional context7 and obscura MCP servers and the optional i-have-adhd and ponytail plugins - and turn the working rules or the review gate on or off, after showing exactly what will change.

Install / Use

npx skills add bothuany/claude-starter --skill setup

Installs into whichever agent you are using.

About this skill
📄

SKILL.md

Installable skill definition

Quality Score

62/100

Supported Platforms

Claude Code

Our assessment of setup

setup scores 62/100 on our quality scale, 4093rd of 4,704 Development & Engineering skills we index.

Its SKILL.md is 6.4 KB long, split into 5 sections with 1 code example: a thorough specification that gives an agent plenty to work with.

It has no GitHub stars yet, so there is no community track record; judge it on its content.

Substance
29/30
Structure
15/20
Description
15/15
Adoption
0/20
Freshness
5/15

Maintenance, license and trust

  • We could not determine when the repository was last updated.
  • No license is declared. By default that means all rights are reserved: you can read it, but reusing or redistributing it is not clearly permitted. Ask the author before building on it commercially.
  • Its trust signals score 68/100, with 3 cautions from licensing, adoption, age or documentation. These come from repository metadata, not a code audit — read the skill file before letting an agent act on it.

setup compared with similar skills

All 4 of these similar skills score higher than setup; compare them before choosing.

SkillScoreStarsUpdatedFormat
setup (this skill)by bothuany620—SKILL.md
taste-skillby Leonxlnx100117.0k8d agoCLAUDE.md
Agent-Reachby Panniantong100112.4k9d agoCLAUDE.md
rufloby ruvnet10081.9ktodayCLAUDE.md
headroomby headroomlabs-ai10079.9ktodayCLAUDE.md

Frequently asked questions

How do I install setup?
Run npx skills add bothuany/claude-starter --skill setup. The install tabs above show the steps for each supported agent.
Which AI agents does setup work with?
It is written for Claude Code, as a SKILL.md file. Other agents that read the same format can often use it too.
Is setup safe to use?
It declares no license and scores 68/100 on trust signals. Skills are instructions an agent will follow, so read the file before installing it and do not approve commands you do not understand.
Is setup still maintained?
We could not determine when the repository was last updated.

name: setup description: Configure claude-starter's pieces that a plugin can't ship itself - the optional context7 and obscura MCP servers and the optional i-have-adhd and ponytail plugins - and turn the working rules or the review gate on or off, after showing exactly what will change. Also removes the old copy of the rules that versions before 0.8 wrote into ~/.claude/CLAUDE.md. Run with "check" to only report the current state, or with "skip" / "unskip" to stop or resume the per-project reminders about REVIEW.md and claude-code-setup. disable-model-invocation: true argument-hint: "[check | skip review-md|project-setup | unskip review-md|project-setup]"

The scripts live in ${CLAUDE_PLUGIN_ROOT}/scripts. Run them with node. Nothing here may change a file before the user has seen the change and agreed.

0. Project reminders

At session start the plugin reminds, once a day per project, when the project has no REVIEW.md (/starter:review-setup drafts it) or hasn't run the separate claude-code-setup plugin (/claude-code-setup:claude-automation-recommender; only mentioned when that plugin is installed). If $ARGUMENTS is skip review-md, skip project-setup, unskip review-md or unskip project-setup, run node "${CLAUDE_PLUGIN_ROOT}/scripts/setup.mjs" <skip|unskip> <review-md|project-setup> --data "${CLAUDE_PLUGIN_DATA}", repeat its one-line answer, and stop. skip stops the reminder for the current project only; unskip brings it back.

1. Check

Run node "${CLAUDE_PLUGIN_ROOT}/scripts/setup.mjs" check --data "${CLAUDE_PLUGIN_DATA}" and summarize it in one short table: Node version, rules (on / off), old rules block in CLAUDE.md (none / present / broken), context7 registered, obscura (binary found, MCP registered), i-have-adhd installed, ponytail installed, review gate on or off, and the two project reminders (projectReminders: review-md and project-setup, each unknown / done / declined for this project, plus whether claude-code-setup is installed).

If $ARGUMENTS is check, stop here.

2. Show the plan

  • Rules (on by default). The plugin's SessionStart hook puts ${CLAUDE_PLUGIN_ROOT}/rules.md into every session's context (450 words); nothing is written to CLAUDE.md. Show the text when the user asks or when they are off. If the user's own CLAUDE.md already says the same things, point out the overlap: they may prefer to turn the rules off or trim their own lines.

  • Old rules block. When it is present, the hook holds the rules back so they don't load twice: offer to remove it. Only the marked block goes, the rest of CLAUDE.md stays, and the old file is kept as a timestamped CLAUDE.md.<time>.starter-backup. When it is broken (unbalanced or out-of-order markers), the script won't touch the file: show the user the marker lines and ask them to delete the old block by hand.

  • context7 (optional). An MCP server with current library documentation (upstash/context7), keyless, added at user scope. It sends library names and topics to context7.com: on a work machine, check that this is allowed. Offer it only if no context7 server is registered yet.

  • obscura (optional). A local headless browser with an MCP server (h4ckf0r0day/obscura). If the binary was found, offer to register it at user scope. If not, say how to get it: download the normal build for the OS from its releases page (not a -stealth build: that one is made to evade bot detection) and put it on PATH, then run /starter:setup again. Without it, WebSearch and WebFetch are used.

  • Review gate (on by default). A Stop hook: when this session changed 2+ hand-written code files (15+ lines) or 50+ such lines since the last review and code-review hasn't run, it asks Claude once to run it before stopping. Tests, docs, lockfiles, build output, pure renames and whitespace don't count. It adds a turn and some cost on larger changes; Sonnet skipped the review rule without it. Offer to turn it off only if the user asks or prefers fewer stops; CLAUDE_STARTER_REVIEW_GATE=off in settings env also works.

  • i-have-adhd (optional, third party: ayghri/i-have-adhd, MIT, pinned to v0.4.1). Output shaping: answers lead with the next action, numbered steps, no tangents. It adds its own skill and a SessionStart hook that puts its instructions into every session. Offer it only if it isn't installed.

  • ponytail (optional, third party: DietrichGebert/ponytail, MIT, pinned to v5.3.3). "Lazy senior dev" mode: reuse what the repo already has, standard library first, the smallest code that works, never cutting validation, security or tests. Say plainly what it adds:

    • its instructions in every session, through its own hooks;
    • /ponytail-review and /ponytail-audit commands;
    • a cost and token band above the prompt;
    • after every turn, a row with the task's cost, tokens and tests written to ~/.config/ponytail/band/. That stays local unless the user sets a webhook in its plugin settings.

    The review gate counts only the bundled code-review, so a /ponytail-review alone doesn't satisfy it. Offer ponytail only if it isn't installed from any marketplace.

3. Ask

Use AskUserQuestion, one question per item that would change something (skip items already in place; at most 4 questions in one call, so ask the rest in a second call):

  1. Old rules block: remove (only when present).
  2. context7: register / skip.
  3. obscura: register / skip (only when the binary was found).
  4. i-have-adhd: install / skip.
  5. ponytail: install / skip.
  6. Rules: keep on / turn off (only if the user raised it, or they are off: then turn on / keep off).
  7. Review gate: keep on / turn off (only if the user raised it, or it is currently off: then turn on / keep off).

4. Apply

Build one command with a flag for each answer, and run it when there is at least one flag:

node "${CLAUDE_PLUGIN_ROOT}/scripts/setup.mjs" apply [--remove-old-rules] [--rules on|off] [--context7] [--obscura] [--review-gate on|off] --data "${CLAUDE_PLUGIN_DATA}"

For i-have-adhd, run claude plugin install i-have-adhd@claude-starter. For ponytail, run claude plugin install ponytail@claude-starter; it takes effect in the next session.

Then run check again and report the new state. Say that rule changes take effect in the next session, and that a newly registered MCP server shows up after a restart.

Related Skills

View on GitHub
GitHub Stars0
CategoryDevelopment
UpdatedNaNy ago
Forks0

Trust signals

68/100

From repository metadata: license, adoption, age and documentation. Not a code audit — see the Safety scan above for what the skill file itself contains.

2 medium1 low